Improved Bankfull Discharge Prediction Using 2‐Year Recurrence‐Period Discharge 1

Knowledge of bankfull discharge ( Q bf ) is essential for planners, engineers, geomorphologists, environmentalists, agricultural interests, developments situated on flood prone lands, surface mining and reclamation activities, and others interested in floods and flooding. In conjunction with estimating Q bf , regionalized bankfull hydraulic geometry relationships, which relate Q bf and associated channel dimensions (i.e., width, depth, and cross‐section area) to drainage basin area ( A da ), are often used. This study seeks to improve upon the common practice of predicting Q bf using A da exclusively. Specifically, we hypothesize that predictions of Q bf can be improved by including estimates of the 2‐year recurrence‐period discharge ( Q 2 ) in regression models for predicting Q bf . For testing this hypothesis, we used Q bf estimates from 30 reports containing data for streams that span 34 hydrologic regions in 16 states. Corresponding values of Q 2 and A da were compiled from flood‐frequency reports and other sources. By comparing statistical measures (i.e., root mean squared error, coefficient of determination, and Akaike’s information criterion), we determined that predicting Q bf from Q 2 rather than A da yields consistently better estimates of Q bf . Other principal findings are (1) data are needed for at least 12 sites in a region for reliable hydraulic geometry model selection and (2) an approximate range of values for Q bf / Q 2 is 0.10‐3.0.